Cognitive Liberties
Thought diversity · Executive overview

Executive synthesis — Reclaim cognitive liberty through expression and discoverability

Research edition · October 11, 2026 · 1154 source words

Explore all studiesInspect research sources

Research date: October 11, 2026
Project: https://cognitiveliberties.com
Thesis: Cognitive liberty is the freedom to seek, receive and impart ideas; voluntary participation and access to diverse viewpoints are central, and privacy is an individual option rather than the primary goal.

What must change

The earlier archive overemphasized anonymity, zero-retention, privacy-preserving AI and withdrawal from training datasets. Those safeguards can matter to some people, but they are not the guiding measures of a movement whose aim is that human perspectives remain accessible and honestly represented. The revised research focuses on four public capabilities: inquire without inferred guilt; express and publish; encounter credible disagreement and unfamiliar ideas; and correct or appeal discovery failures.

Findings supported by sources

  1. Human rights: Articles 18–19 of the UDHR protect thought, opinion, and seeking/receiving/imparting information; they do not independently guarantee inclusion or equal rank in private search products. [S01][S24]
  2. Diversity in AI: UNESCO calls for inclusive AI development, fair access and participation across cultures, languages and opinions, providing an important normative basis for a diversity-focused initiative. [S02][S06][S25]
  3. Pluralistic alignment: A July 2026 public-document review found no clear public evidence that production frontier systems explicitly use pluralism as a training/evaluation goal; that does not establish what undisclosed internal programs do. [S03]
  4. Technical correction: Model collapse from recursively training on model-generated data is a different phenomenon from omitting human minority viewpoints. Both are worth researching, not conflating. [S05]
  5. Links: Google encourages normal contextual external links and targets ranking-manipulative link schemes. Linking independent sources is a core editorial service, not a forbidden act. [S07][S08]
  6. AI and source access: Pew's March 2025 browsing panel observed ordinary search clicks in 8% of visits with a Google AI summary versus 15% without. The analysis is observational; it raises questions about source visits without proving intentional suppression. [S09]
  7. Website correction: The site's Resource Navigator, network, organizations directory, standards directory and research evidence ledger are accessible in live public checks. An earlier supplemental document's “dead-end” claims are false as of these checks. [S19][S20][S21][S22][S23]
  8. Technical distinction: GEO/AEO concern content visibility and answer usability; AG-UI is an event protocol for agent/user interaction, not an SEO tactic, training permission or evidence of pluralism. [S10][S16][S17]

Proposed central doctrine

Think freely. Search without prejudgment. Share what you know. Find the dissent. Inspect the evidence. Challenge exclusion.

This is a normative editorial doctrine; it should be debated openly. A pluralistic platform must not equate a reference to a viewpoint with endorsement, and it must distinguish evidence-based conclusions from social preferences, speculation and coordinated manipulation.

Specific changes to the existing website

The public homepage currently foregrounds “Your mind is your own. Your questions should be, too” and emphasizes private inquiry. It also warns that ranking can make lawful publishers effectively invisible. Recenter the homepage on the latter insight and the Article 19 triad: seek, receive, impart. Preserve existing standards and private-inquiry pages for people who need them, but stop using confidentiality as the organizing story. [S18]

The existing navigator already offers eight starting questions. Add pathways about finding opposing perspectives, voluntarily publishing viewpoints, checking whether a source was omitted, contributing research and requesting public corrections. Link these to its existing network and evidence ledger rather than creating an entirely new repository. [S19][S20][S21]

First 30 days: mission and home-page rewrite; publish the positive cognitive-liberty charter and this corrections ledger; choose three diverse topic trails.

Days 31–90: publish annotated sources, invite voluntary contributions, implement an editorial challenge process and test source-discovery tasks with consenting participants.

Days 91–180: run a reproducible AI/search viewpoint-coverage audit; pilot a cross-publisher voluntary discovery federation.

Days 181–365: publish independent results, expand multilingual coverage, and consider an optional AG-UI link navigator only if a static version does not meet navigation goals.

Measures of success

Use successful original-source discovery; fidelity of viewpoint descriptions; availability of rebuttals; language and independent-author coverage; omission correction time; documented contribution reuse; navigation task completion. Do not measure success by how little knowledge is shared, how little AI can learn, or by unverifiable promises that every opinion will rank equally.

Essential counterargument

Maximizing the number of visible perspectives without quality controls can enable spam and manufactured controversy. The goal is freedom to access and evaluate—not mandatory endorsement, equal scientific credibility, or guaranteed model training. Test all proposals against both viewpoint exclusion and misleading false balance.

What this research does not claim

No internal site audit, personal visitor tracking, live AI product benchmark, platform source-code inspection, training corpus inspection or controlled field trial was performed. The sample app is an offline demonstration only. Conclusions about influence on deployed ranking systems remain hypotheses to evaluate.

References

Continue independent inquiry

Follow the cited primary sources, compare their evidence, and draw your own conclusions. Disagreement is part of discovery—not evidence of wrongdoing.

Browse the full research collection · Follow the resource navigator · Review corrections